Strange voltage/temperature response

dkm777

Senior member
Nov 21, 2010
528
0
0
Hi guys,

During my overclocking of my i7 I noticed that certain lower voltages would give me higher temps than higher voltages. Does anybody have an idea of why this is happening? I have an unfounded theory that the lower voltage is right on the edge of stability and the ECC hardware for L2 and L3 caches is working hard correcting little errors.
 

Soulkeeper

Diamond Member
Nov 23, 2001
6,739
156
106
hmmm
I suppose your theory is possible, but more likely you are getting false temperature readings.
Where are you reading your temperature ?
are the power saving features enabled in the bios (ie: speedstep) ?
 

dkm777

Senior member
Nov 21, 2010
528
0
0
I have all the features enabled and we're talking a small, but consistent difference of ~3 degrees as reported by HWMonitor.
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,081
3,583
126
u cant assume this unless EVERY voltage u have is hard set... meaning u dont have ANYTHING on auto.

For all you know it could just be the Northbridge generating more heat... or the south bridge generating more heat.. or your cant finding a nice spot ontop of your PC and resting there.. :p
 

watek

Senior member
Apr 21, 2004
937
0
71
u cant assume this unless EVERY voltage u have is hard set... meaning u dont have ANYTHING on auto.

For all you know it could just be the Northbridge generating more heat... or the south bridge generating more heat.. or your cant finding a nice spot ontop of your PC and resting there.. :p

PC butt warmer? Ha
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
I have all the features enabled and we're talking a small, but consistent difference of ~3 degrees as reported by HWMonitor.

Yeah that is strange. The physics of power consumption and voltage are pretty straight forward and incontrovertible, so if the observation is real and repeatable then it must be something to do with a PLL or VRM somewhere in the loop that is operating at the near end of a voltage range (so its efficiency is lower) at the lower voltage but when you step up the voltage one more notch the VRM/PLL changes modes and starts operating in a more efficient portion of its power curve?

An analogy to the ram straps on FSB systems like LGA775 is about the best I can mentally visualize.

attachment.php
 
Last edited: